Habr AI
AI news source. Articles are auto-selected and adapted by Hamidun News editors.
Latest publications
Authentication in the MCP protocol in 2026: past, present, and future
An analysis of MCP protocol evolution from Anthropic—from local calls like 'check the calendar' to remote servers where authentication became mandatory.
AI regulation in Russia in July 2026: Digital Ministry, law on digital financial markets and Estonian IDs for agents
A Habr author who has been observing Russia's AI policy since November 2025 calls July 2026 a turning point for AI regulation in Russia.
Five AI plugins for Obsidian that help retain knowledge from notes
A developer from IBS shares five AI plugins for Obsidian that help search for information by meaning, communicate with a knowledge base and process articles and meetings faster.
Neuroanalytic 2.0 in Yandex DataLens: overview of an AI agent for data analysis
Yandex introduced Neuroanalytic 2.0 — an updated AI assistant in the BI system DataLens, which now works as a full-fledged AI agent rather than just a data assistant.
How to evaluate AI agent libraries: the example of transformers and traces over answers
Developers of AI agent libraries are moving from checking the correctness of the final answer to analyzing the agent's path itself — traces, tokens, time and errors, using the transformers library as an example.
Why traditional PRD doesn't save AI features in 2026: what to add to the spec
Product specialists discuss what sections should be added to an AI product specification so a feature doesn't break before release due to unpredictable LLM behavior.
Anthropic found an analog of global workspace in Claude — J-space
Anthropic published research on July 6, 2026 on J-space — a structure in Claude similar to the brain's global workspace, but composed of words.
Context engineering: why AI agents get confused in long dialogues and how to fix it
An overview of a new field in AI agent development — context engineering, which solves the problem of LLM response degradation in long scenarios with RAG, tools and memory.
OpenAI releases GPT-5.6: flagship Sol, Terra model and budget-friendly Luna
OpenAI moved the GPT-5.6 family from limited testing to general availability — flagship Sol, balanced Terra and budget-friendly Luna, plus a new ultra mode with parallel agents.
Photonic AI Chip Computes at Light Speed: 99% Accuracy Without Overheating
Sydney University scientists created a nanophotonic processor that processes data with light instead of electricity in picoseconds.
Neural network gives Sberbank 67% chance to win 1.44 billion ruble court case
neshemyaka service used AI to predict the outcome of a cassation case between FIT and Sberbank.
Sora closed after 4 months: collapse of OpenAI and Disney's $1 billion video generator
OpenAI's video generator Sora was announced in December 2025 and officially closed on March 24, 2026, leaving billions in losses and questions about the future of video AI.
How Rosatom purchased AI tools through unofficial channels
A story about how saving on a professional SMM team led a Russian company to buying AI and tech resources through unofficial channels, risking its reputation.
How to Use Free AI for Coding: A Realistic Guide
Overview of available free AI tools for coding: capabilities, limitations, and when to upgrade.
Simple Diffusion: compact model 100 times cheaper than SDXL
A team created a diffusion model sdxs-1b (training $600K vs SDXL $6M), runs on consumer GPUs with quality close to SDXL.
Claude Code Leak Reveals: It's Not an API Wrapper, but a Full Operating System
512 thousand lines of Claude Code source code exposed the architecture Anthropic was hiding: self-healing loops, computer vision, two-tier function filtering.
Anthropic and AI Safety Marketing: A Trillion-Dollar Empty Slogan?
Critics on Habr dissect Anthropic's claims of AI safety leadership as marketing for a multi-billion-dollar industry.
RPA and AI agents in corporate architecture: not instead, but together
RPA and LLM agents solve different tasks: the former for strictly structured processes, the latter for unstructured data. In enterprise this is two layers of one system.
How to make English-language neural networks read in Russian
Review of free Text-to-Speech models that voiceover text without credits and subscriptions, focusing on Russian language support and synthesis quality.
OpenClaw skills: installation, creation, and protection against 341 malicious sets
On the OpenClaw platform, 341 malicious skills were discovered among 2,857 verified — these are instructions by which an agent reads files, runs commands, and accesses the network without permission.
ChatGPT will sketch your presentation, but you'll do the redesign
Professional presentation designer shares experience using ChatGPT in workflow: AI helps at different stages, but leaves critical work to humans.
PocketCoder-A1: how to make Claude work in three shifts
Habr published case about developer who created automatic coder continuously working with Claude API to maximize project output from cloud subscription.
Can AI replace package managers? From npm to a prompt registry
Vision of the future of development: instead of package managers like npm, developers publish prompts, and AI generates code on the fly, avoiding dependencies and supply chain attacks.
AI Application in Manufacturing: 6 Real Implementation Examples
Overview of six practical examples of implementing AI technologies on manufacturing enterprises with proven economic effect.
Bounding Box Augmentation in Detection: Formats, Errors, and Best Practices
Technical guide on managing bounding box coordinates during image augmentation in computer vision, including common Albumentations configuration errors.
Running GLM-5.1 Locally: Complete Guide to llama.cpp and GGUF
Habr published a detailed guide for deploying GLM-5.1 model on local hardware using llama.cpp tools and GGUF formats for memory optimization.
Can an AI agent be a network engineer—reset a Cisco router password
An engineer tested whether an LLM agent can handle the 'black magic' of networking: COM-port connection, ROMMON work, timings. Answer: it can.
Multi-Agent AI Platforms for Retail: End of Loneliness
GlowByte presented an article on how multi-agent AI systems transform retail, helping category managers in making multi-million ruble margin decisions.
Habr: Comparison of AI Models in Developing Battle City Remake (2026)
Russian developer community repeated experiment comparing various AI models when creating game application, showing progress in model capabilities over six months.
FINESSE-Bench: Evaluating LLMs in Financial Domain Without Deception
Finam AI Lab updated the open FINESSE-Bench to verify whether LLMs can truly work with financial data and trading tasks.